• The KIPS Transactions:PartB
• /
• v.10B no.3
• /
• pp.281-286
• /
• 2003

A large number of features are collected for problem solving in real life, but to utilize ail the features collected would be difficult. It is not so easy to collect of correct data about all features. In case it takes advantage of all collected data to learn, complicated learning model is created and good performance result can't get. Also exist interrelationships or hierarchical relations among the features. We can reduce feature's number analyzing relation among the features using heuristic knowledge or statistical method. Heuristic technique refers to learning through repetitive trial and errors and experience. Experts can approach to relevant problem domain through opinion collection process by experience. These properties can be utilized to reduce the number of feature used in learning. Experts generate a new feature (highly abstract) using raw data. This paper describes machine learning model that reduce the number of features used in learning using heuristic function and use abstracted feature by neural network's input value. We have applied this model to the win/lose prediction in pro-baseball games. The result shows the model mixing two techniques not only reduces the complexity of the neural network model but also significantly improves the classification accuracy than when neural network and heuristic model are used separately.

• The Korean Journal of Archival Studies
• /
• no.59
• /
• pp.207-275
• /
• 2019

This archiving project of the survivors of suicide was done with the survivor supporting team of the Seoul Suicide Prevention Center. The survivor supporting team was operating a Self-help Support Group for the emotional support of the survivors of suicide. A Self-help Support Group is a place for the survivors of suicide to regularly meet and share their suffering by talking of topics hard to discuss elsewhere. As the Self-help Support Group progressed members who acted as the leader of the group appeared. They formed an essay group that writes together. Two fathers who lost their sons, two mothers who lost their daughters, a mother who lost her son, a wife who lost his husband. The essay group met each week in a place facing Sajik Park. Through the windows that took up the whole side of the room, evening was coming in. The things that happened during the day went away towards Inwang mountain following the setting sun. Ten people (six members of the essay group, three from the survivor support team, a historian for unique conversation) sat around a table, facing each other. "Now, what shall we do?" History for unique conversation is a time that archives life by sharing conversations. At times a complete stranger, and other times people who share their ordinary lives sit around together (3-9 people, sometimes about 15). On the table there is coffee, bread, fruits and salads, and sometimes a dish someone heartily prepared. When a bottle of wine is placed on the table, each takes a glass. Morning, afternoon, the time the evening is welcomed in, late night. It does not matter which. For six months, 3 hours when meeting every week, 6 hours when at every other week. A room where the ambience is like that of a kitchen where sunlight enters, or a cozy living room is the best location. However, there are many times when it is held in a multipurpose room in the suburbs where many meetings are held, or in a classroom of a school. The meeting place is decided according to different situations of the time. There are no participation requirements as it is said to be for themselves to write down according to archiving form while looking back their lives thoroughly, and they are the only ones to stop themselves. The archives landscape from far away would seem like trying to do some talking. However, when going into a microscopic situation one must leave themselves to the emotional dynamics. It is because it archives the frustration and failures one experienced through life. A participator of history for unique conversation must face the sufferings of their life. The archiving project took place in 2013 to 2014. Many years have passed. Has the objective distance for archiving the situation of that time been secured? That may be uncertain, but I will speak of a few stray thoughts on archiving while depicting the process and method of operation.

• The Korean Journal of Archival Studies
• /
• no.33
• /
• pp.85-112
• /
• 2012

Group for the People without History(GPH) was launched at September 2002 and had worked for around five years with the following purposes; Firstly, GPH collects first-hand data on people's everyday lives based on fieldworks. Secondly, GPH constructs digital archives of the collected data. Thirdly, GPH guarantees the accessibility to the archives for people. And lastly, GPH promotes users to utilize the archived data for the various levels. GPH has influenced on the construction of archives on everyday life history as well as the research areas such as anthropology and social history. What is important is that GPH tried to construct digital archives even before the awareness on archives was not widely spreaded in Korea other than formal sectors. Furthermore, the GPH archives proposed a model of open archives which encouraged the people's participation in and utilization of the archives. GPH also showed the ways in which archived data were used. It had published forty seven books of people's life histories and five photographic books, and held six photographic exhibitions on the basis of the archived data. Though GPH archives had contributed to the ignition of the discussions on archives in various areas as leading civilian archives, it has a few limitations. The most important problem is that the data are vanishing too fast for researchers to collect. It is impossible for researchers to collect the whole data. Secondly, the physical space and hardware for the data storage should be ensured. One of the alternatives to solve the problems revealed in the works of GPH is to construct community archives. Community archives are decentralized archives run by people themselves to preserve their own voices and history. It will guarantee the democratization of archives.

• The Korean Journal of Archival Studies
• /
• no.72
• /
• pp.91-163
• /
• 2022

The focus of this paper lies on the fact that the 'management' and 'utilization' of audiovisual archives are still not specialized in both the public and the private sectors. The use of online video platforms including 'YouTube' has became common. Accordingly the production and collection of high-definition and high-capacity audiovisual archives has been rapidly increasing. However, it also emphasizes that there are no references or principles in the current Public Records Act and its enforcement rules, public standards, and guidelines. This paper ultimately examines the provisions that are related to audiovisual archives of the current Public Records Act, which needed to be revised and enacted due to the lack of an audiovisual archives management manual of national institutions, public broadcasters, and organizations can refer to. In addition, this study tries to find out what kind of systems and guidelines are used in audiovisual archives management. This paper examines the current state of standardization of audiovisual records of the National Archives. It also analyses the systems and the guidelines methodically for efficient audiovisual record management in the public records management sector. It suggests the new direction of relevant public standards and guidelines through this research. Futhermore, it measures to activate the audiovisual management policy-making functions of the National Archives. The necessity of establishing a Public Audiovisual Archives as an organization was also reviewed in this paper. The Public Audiovisual Archives will collect Public Audio and Videos systematically and comprehensively through the legal deposit system. And it will be operated by the management and the utilization system so that it can be used for public as a collective memory. Finally, it will takes a charge of a professional role in audiovisual record management field, such as technology standardization to safeguard and protect the copyrights through this process.

• The Bulletin of The Korean Astronomical Society
• /
• v.46 no.2
• /
• pp.73.4-74
• /
• 2021

한국천문연구원은 2017년 제1차 구술채록사업에 이어 2020년 제2차 사업을 진행하면서 최초로 원외 원로에 대한 구술채록을 시도하였다. 국가 대표 천문연구의 산실로서 연구원 존재 의의를 확립하기 위하여 원내 원로에 국한되었던 구술자 대상을 확장한 것이다. 그 첫 외부 구술 대상자로 방득룡 전임 노스웨스턴 천문학과 교수를 선정하여 2020년 7월부터 준비단계에 들어갔다. 방득룡 전(前)교수가 첫 번째 한국천문연구원 원외 인사 구술자로 선정된 이유는, 그가 우리나라 천문대1호 망원경 구매 선정에 개입한 서신(1972년)이 자료로 남아있었기 때문이다. 한국천문연구원에서 2017년에 수행한 제1차 구술채록사업에서 구술자로 참여한 오병렬 한국천문연구원 원로가 기증한 사료들은 대부분 연구원 태동기 국립천문대 구축과 망원경 구매 관련 자료였으며 이 가운데 1972년 당시 과학기술처 김선길 진흥국장에게 Boller and Chivesns(사(社))의 반사경을 추천한 방득룡 전(前)교수의 서신은 한국 천문학 발전사에서 중요한 사료였다. 연구진은 이 자료를 시작으로, 방득룡 전(前)교수의 생존 여부와 문서고의 공기록물들에서 그의 흔적을 찾아가기 시작했다. 놀랍게도 그는 실제 세계와 한국천문연구원 문서고 깊숙이 기록물들 모두에서 상존하고 있었다. 1927년생인 방득룡 전(前)교수, Dr. John D. R.은 미국 플로리다 한 실버타운에서 건강한 정신으로 생존하여 있었고 연구진의 인터뷰에 흔쾌히 응했다. 2020년 9월 16일에 한국천문연구원 본원 세종홀 2층 회의실에서 영상통신회의로 그와의 구술인터뷰가 진행되었다. 이 구술인터뷰는 원외 인사가 대상이란 점 외에도 방법적으로는 전형적인 대면 방식이 아닌 영상 인터뷰였다는 점에서 코로나 시대의 대안이 되는 실험적 시도였다. 현대 한국천문학 발전사의 재조명 측면에서도 의미가 있었다. 1960년대 초반부터 1992년 정년퇴임까지 30년을 미국 유수 대학교 천문학과 교수로 재직하며 활발한 활동을 해 온 한국계 천문학자가 우리나라 최초 반사망원경 구매 선정에 적극 개입하였던 역사는, 공문서 자료들과 서신 사료들에 이어 그의 육성으로 나머지 의구심의 간극이 채워졌다. 또 구술자 개인이 주관적으로 중요하다고 여기는 '기억'이 중요한 아카이빙 콘텐츠 확장의 단초가 될 수 있다는 것을 보여줌으로써 구술사 연구에 있어서도 중요한 관점을 주었다. 애초 연구진이 방득룡 전(前)교수의 공식 기록에서 아카이빙의 큰 줄기로 잡았던 것은 1948년 도미, 1957년 위스콘신 대학교 천문학 박사학위 취득, 1962년부터 노스웨스턴 대학(일리노이주 에반스턴)의 천문학 교수진, 1992년 은퇴로 이어진 생애였다. 그러나 그와의 구술 준비 서신 왕래와 구술을 통하여 알게 된 그가 인생에서 중요시 여겼던 지점은, 1948년 도미 무렵 한국의 전쟁 전 상황과 당시 비슷한 시기에 유학한 한국 천문학자들의 동태, 그리고 1957년부터 1962년까지 프린스턴 대학교에서 M. Schwarzschild 교수와 L. Spitzer 교수를 보조하며 Stratoscope Project를 연구하였던 경험이었다. 기록학적 의미에서도, 전자를 통해서 그와 함께 동시대 한국 천문학을 이끌었던 인재들의 맥락정보를 얻을 수 있었으며, 후자를 통해서는 세계 천문학사에 큰 영향을 미친 석학에 대한 아카이브 정보와의 연계 지점과 방득룡 전(前)교수의 연구 근원을 찾을 수 있었다. 이들은 추후 방득룡 콘텐츠 서비스 시에 AIP, NASM, Lyman Spitzer 콘텐츠, 평양천문대, 화천조경천문대, 서울대와 연세대, 그리고 한국천문연구원까지 연계되어 전 세계 폭넓은 이용자들의 유입을 유도할 수 있는 검색 도구가 될 수 있다. 이번 방득룡 구술사 연구에서 구술자 개인의 주관적인 소회가 공식 기록이 다가갈 수 없는 역사적 실체에 일정 부분 가까울 수 있다는 것, 그리고 이를 통하여 개인의 역사는 공동체의 역사로 확장될 수 있다는 사실을 발견할 수 있었다. 또 연구진은 방득룡 전(前)교수의 회상을 통하여 구술자 개인의 시각으로 한국과 미국 천문학계의 공동체 역사를 재조명할 수 있었고, 이것을 아카이브 콘텐츠 확장 서비스에 반영할 수 있다는 기대를 가지게 되었다. 무엇보다 이 연구를 통하여 다양한 주제의 아카이브로 연동될 수 있는 주제어와 검색도구를 구술자 개인의 회상으로부터 유효하게 도출할 수 있다는 것을 확인하였다. 그리고 향후 한국천문 구술아카이브의 확장을 통하여 보다 다양한 활용과 연구 재활용의 선순환이 가능하다는 것도 알 수 있었다. 이는 최근 기록학계에서 대두되고 있는 LOD(Linked Open Data)의 방향성과도 흡사하여 한국천문학 구술사연구의 차세대 통합형 기록관리의 미래모형을 기대케 하는 대목이다.

• Journal of Korea Entertainment Industry Association
• /
• v.13 no.8
• /
• pp.313-323
• /
• 2019

For the actors on video, facial expression acting can easily become 'forced facial expression' or 'over-acting'. Also, if self-restraint is emphasized too much, then it becomes 'flat acting' with insufficient emotions. By bringing forth questions in regard to such facial expression acting methods, this study analyzed the facial expression acting of the actors in genre dramas with strong commercial aspects. In conclusion, the facial expression acting methods of the actors in genre dramas were being conducted in a typical way. This means that in visual conventions of video acting, the aesthetic standard has become the important standard in the facial expression acting of the actors. In genre dramas, the emotions of the characters are often revealed in close-up shots. Within the close-up shot, the most important expressive medium in a 'zoomed-in face' is the 'pupil of the eye', and emotions are mostly expressed through the movements of the eye and muscles around it. The second most important expressive medium is the 'mouth'. The differences in the degree of opening and closing the mouth convey diverse emotions along with the expression of the 'eye'. In addition, tensions in the facial muscles greatly hinder the expression of emotions, and the movement of facial muscles must be minimized to prevent excessive wrinkles from forming on the surface of the face. Facial expressions are not completed just with the movement of the muscles. Ultimately, the movement of the muscle is the result of emotions. Facial expression acting takes place after having emotional feelings. For this, the actor needs to go through the process of 'personalization' of a character, such as 'emotional memory', 'concentration' and 'relaxation' which are psychological acting techniques of Stanislavsky. Also, the characteristics of close-up shots that visually reveal the 'inner world' should be recognized. In addition, it was discovered that the facial expression acting is the reaction acting that provides the important point in the unfolding of narratives, and that the method of facial expression and the size of the shots required for the actors are different depending on the roles of main and supporting characters.

• Journal of Intelligence and Information Systems
• /
• v.23 no.1
• /
• pp.95-108
• /
• 2017

Recently, AlphaGo which is Bakuk (Go) artificial intelligence program by Google DeepMind, had a huge victory against Lee Sedol. Many people thought that machines would not be able to win a man in Go games because the number of paths to make a one move is more than the number of atoms in the universe unlike chess, but the result was the opposite to what people predicted. After the match, artificial intelligence technology was focused as a core technology of the fourth industrial revolution and attracted attentions from various application domains. Especially, deep learning technique have been attracted as a core artificial intelligence technology used in the AlphaGo algorithm. The deep learning technique is already being applied to many problems. Especially, it shows good performance in image recognition field. In addition, it shows good performance in high dimensional data area such as voice, image and natural language, which was difficult to get good performance using existing machine learning techniques. However, in contrast, it is difficult to find deep leaning researches on traditional business data and structured data analysis. In this study, we tried to find out whether the deep learning techniques have been studied so far can be used not only for the recognition of high dimensional data but also for the binary classification problem of traditional business data analysis such as customer churn analysis, marketing response prediction, and default prediction. And we compare the performance of the deep learning techniques with that of traditional artificial neural network models. The experimental data in the paper is the telemarketing response data of a bank in Portugal. It has input variables such as age, occupation, loan status, and the number of previous telemarketing and has a binary target variable that records whether the customer intends to open an account or not. In this study, to evaluate the possibility of utilization of deep learning algorithms and techniques in binary classification problem, we compared the performance of various models using CNN, LSTM algorithm and dropout, which are widely used algorithms and techniques in deep learning, with that of MLP models which is a traditional artificial neural network model. However, since all the network design alternatives can not be tested due to the nature of the artificial neural network, the experiment was conducted based on restricted settings on the number of hidden layers, the number of neurons in the hidden layer, the number of output data (filters), and the application conditions of the dropout technique. The F1 Score was used to evaluate the performance of models to show how well the models work to classify the interesting class instead of the overall accuracy. The detail methods for applying each deep learning technique in the experiment is as follows. The CNN algorithm is a method that reads adjacent values from a specific value and recognizes the features, but it does not matter how close the distance of each business data field is because each field is usually independent. In this experiment, we set the filter size of the CNN algorithm as the number of fields to learn the whole characteristics of the data at once, and added a hidden layer to make decision based on the additional features. For the model having two LSTM layers, the input direction of the second layer is put in reversed position with first layer in order to reduce the influence from the position of each field. In the case of the dropout technique, we set the neurons to disappear with a probability of 0.5 for each hidden layer. The experimental results show that the predicted model with the highest F1 score was the CNN model using the dropout technique, and the next best model was the MLP model with two hidden layers using the dropout technique. In this study, we were able to get some findings as the experiment had proceeded. First, models using dropout techniques have a slightly more conservative prediction than those without dropout techniques, and it generally shows better performance in classification. Second, CNN models show better classification performance than MLP models. This is interesting because it has shown good performance in binary classification problems which it rarely have been applied to, as well as in the fields where it's effectiveness has been proven. Third, the LSTM algorithm seems to be unsuitable for binary classification problems because the training time is too long compared to the performance improvement. From these results, we can confirm that some of the deep learning algorithms can be applied to solve business binary classification problems.